Modeling Ground Ozone Concentration Changes after Variations in Precursor Emissions and Assessing Their Benefits in the Kanto Region of Japan

Abstract

Ozone (O₃) is a pollutant of concern in urban areas because of its effects on health, crops, ecosystems, and materials. Despite efforts to meet the Japanese air quality standard for O₃ in the Kanto region, the attainment percentage is close to zero. Considering that O₃ formation is sensitive to emissions of volatile organic compounds (VOC) and nitrogen oxides (NO_x), this study evaluated a range of reductions in the emissions of both precursors using a regional air quality model (ADMER-PRO) and estimated their benefits measured as the economic change due to O₃ concentration differences between scenarios. The simulation period was set during the 2016 O₃ season. The results showed that O₃ concentrations could be reduced using two approaches: significant reduction in VOC levels combined with minor NO_x level changes or significant NO_x emission reduction. Significant reduction in NO_x levels was the most effective strategy for a generalized decrease in the O₃ levels in the Kanto region, and the benefit analysis revealed that the most significant economic impacts could be achieved by adopting the latter approach.

1. Introduction

Air pollution remains one of the greatest environmental threats to health, and the related global burden of disease has increased in the last 25 years because of the aging population, increased non-communicable disease rates, and deterioration of air quality in low- and middle-income countries [1]. Among the wide variety of air pollutants, particles with an aerodynamic diameter <2.5 µm (PM_2.5) and tropospheric ozone (O₃) are the two species most frequently used to quantify the adverse effects of air pollution exposure [2].

Although the main issues with PM_2.5 are frequently observed in developing nations because of the lack of adequate regulations against it [3,4], O₃ remains a problem in urban and suburban areas of both the developed and the developing world [5]. The air quality guidelines established by the World Health Organization (WHO) for O₃ are far from being attained, even in cities that have strengthened regulations aiming to reduce their levels. The nonlinear relationship between precursor emissions and the amount of O₃ formed is one of the reasons why countermeasures are frequently ineffective. O₃ is a secondary air pollutant that forms in the troposphere after a series of chemical reactions. Emissions of nitrogen oxides (NO_x), volatile organic compounds (VOC), and U.V. radiation play significant roles in its formation. However, depending on the chemical regime, O₃ formation is governed by NO_x emissions (NO_x limited) or VOC emissions (NO_x saturated). This relationship must be considered when addressing ozone pollution [6].

In recent years, the concentrations of air pollutants in Japan, including O₃ precursors, have gradually decreased. However, the levels of ozone remain unchanged despite the establishment of policies to reduce their precursors [7]. In the Kanto region of Japan, the number of days when the air quality standard (AQS) for photochemical oxidants (60 ppb) is attained is close to zero. Moreover, in 2017 there were eighty-seven days where the concentrations of O₃ were double the recommended values of the World Health Organization (100 μg/m³, 8-h daily maximum), which led to the declaration of environmental alerts due to O₃ pollution. The problem worsens during the O₃ season, which spans from April to September, when the ideal conditions of high U.V. radiation, stable weather conditions, and precursor emissions make the perfect mix for generating high O₃ concentrations [8].

The problems associated with O₃ include its adverse effects on health, crops, ecosystems, and materials. Long-term exposure to high O₃ concentrations has been proven to negatively affect health and decrease the life expectancy of exposed people. Additionally, O₃ is an important climate forcer, whose contribution to climate change is as significant as that of methane [9,10,11,12]. Thus, reducing O₃ levels in Kanto would benefit air quality, health, and crop production and contribute to achieving the net-zero emission goal by 2050 [13,14,15].

Ozone is the air pollutant with the most significant adverse effects on plants owing to its phytotoxicity. The types of damage that O₃ causes to plants range from visible effects to physiological changes and alterations in photosynthesis, which are directly related to crop yields [16]. Rice is one of the most susceptible crops to high ozone concentrations. In Asia the production of different crops, including rice, shows reductions from 5% to 29% due to increasing concentrations of ground-level ozone [17]. In Japan, the reductions reported in the rice yield from 2000 to 2005 represented an economic loss of more than 280 million (int. $) [18]. Hence, it is fundamental to consider the effects of ozone on the most important crop for Japan.

Different studies have used air quality models to evaluate the influence of various factors on the levels of O₃ formed, such as meteorology [19,20,21,22] and emissions of precursors from specific sources [23], to address the O₃ issue in the Kanto Plain. Nonetheless, no studies related to the variations in the emission rates of the main precursors have been reported. This approach has been used in other regions to elucidate the best strategies for mitigating tropospheric O₃ levels [24,25,26,27]. Considering this, this study evaluated the O₃ responses to a range of reductions in the emissions of precursors, aiming to generate valuable information on the best approach for achieving changes in O₃ concentrations. The proposed reductions analyze goals that could be achieved in the short term or those that could take longer to be implemented. Furthermore, the economic impacts associated with changes in mortality rates and rice production due to O₃ variations were evaluated. Therefore, the findings provide valuable information for identifying the best strategies to address the O₃ issue in Kanto by delivering an integrated analysis that could be helpful for the enactment of future policies.

2. Materials and Methods

2.1. Study Area

The Kanto Plain is located in the central area of Honshu, the main island of the Japanese archipelago. The region encompasses the Greater Tokyo Area, the most extensive urban area of Japan, formed by the city of Tokyo and Saitama, Kanagawa, and Chiba prefectures. In addition, Ibaraki, Gunma, and Tochigi are part of Kanto in seven prefectures. Most of the land extension is flat terrain with an altitude of <70 m above sea level, except for the western area, where there is a mountainous region with altitudes reaching 2450 m, and the mountains in the southeastern and northeastern areas have maximum heights of 800 m. The annual average temperature is 15.8 °C, the average precipitation is 1598.2 mm, and the relative humidity is 65%.

According to the Statistics Bureau of Japan, the population of Kanto is more than 40 million (Ministry of Internal Affairs and Communications). Along with high population density, the region is characterized by high emissions of air pollutants due to increased industrial activity and vehicle fleets. This study focuses on Kanto, and the capital city of each prefecture was selected for observed-modeled data comparison and specific point analysis (Figure 1).

2.2. Modeled Emissions Scenarios

This study evaluated a range of reductions in total NO_x and VOC emissions from anthropogenic sources and assessed their impact on ground O₃ concentrations. The simulated scenarios considered a combination of cuts of 100%, 50%, and 0% in the VOC emissions with 100%, 90%, 75%, 50%, 25%, 10%, and 0% reductions in the NO_x emissions, for a total of 21 scenarios, including the base year (N100V100).

Table 1. Simulated scenarios with the applied reductions in the nitrogen oxide (NO_x) and volatile organic compound (VOC) emissions.

Scenario	Reduction (%)
	Total NOx	Total VOC
BASE	0	0
N100V050	0	50
N100V000	0	100
N090V100	10	0
N090V050	10	50
N090V000	10	100
N075V100	25	0
N075V050	25	50
N075V000	25	100
N050V100	50	0
N050V050	50	50
N050V100	50	100
N025V100	75	0
N025V050	75	50
N025V000	75	100
N010V100	90	0
N010V050	90	50
N010V000	90	100
N000V100	100	0
N000V050	100	50
N000V100	100	100

provides a description of the reduction applied in each case. The simulated scenarios provided valuable insights into the most effective way to reduce O₃ levels by evaluating targets that could be achieved over the short- and long-term.

2.3. Model Overview

The atmospheric dispersion model for exposure and risk assessment version 1.0 (ADMER-PRO) was used in the simulation. The ADMER-PRO was developed by the National Institute of Advanced Industrial Science and Technology of Japan [24] and has been widely used in air quality research [28,29,30]. The development of ADMER-Pro was based on the regional atmospheric modeling system (RAMS). RAMS uses observed geographic and meteorological data (elevation terrain, sea surface temperature, land use, and temperature) from the National Center for Environmental Prediction (NCEP) in the U.S. to simulate and forecast meteorological phenomena [31]. The generated meteorological data had a resolution of 6 h.

ADMER-PRO couples a chemical transport model to RAMS to perform air quality simulations. The chemical transport model considers atmospheric conditions, emission rates, chemical reactions, and deposition processes for solving mass-balance equations to simulate the concentrations of trace gas species in a specific domain. ADMER-Pro uses the CB99 chemical mechanism. CB99 is an updated version of the Carbon Bond IV Mechanism (CB4) [32] which is a condensed mechanism that considers 37 chemical species and their reactions with the O.H. radicals. The initial and boundary conditions for O₃ were generated by interpolating the observational data from an O₃ sonde located in Tsukuba, Japan. ADMER-PRO uses the methods established by Zhang et al. [33] for the deposition process.

2.4. Emissions Dataset

ADMER-PRO has a built-in emission inventory based on EAGrid2000-Japan [34]. This emission inventory allocated the emitted species to a mesh of 1 km × 1 km with a temporal resolution of 1 h. In our study, the emission datasets were updated from the EAGrid2005 version to 2016 using the method established by Fukui et al. [35].

2.5. Model Set Up

To evaluate the long-term average concentrations of air pollutants, ADMER-PRO features a weather pattern classification analysis [36] that categorizes the types of weather over a period and counts the occurrence of every generated category. It is then possible to set up the simulation time for days with the most representative weather (those with higher occurrence frequency) within a certain period. This study determined the most representative days for the O₃ season in 2016 (occurrence frequency of 24%), finding six days distributed between April and September (Table S1). The simulations were performed over these six days and were considered representative of the O₃ season in 2016. The spin-up period for each simulated day was 51 h.

2.6. Simulation Domains

The simulations were divided into two nested domains (Figure S1). The parent domain covers a significant part of Honshu Island and has a 20 km² resolution. The smaller domain covers the Kanto region, with a resolution of 5 km². The vertical grid configuration considered 29 layers, with the top level at an altitude of 20 km and the bottom layer at an altitude of 50 m.

2.7. Model-Derived and Observed Data Comparison

Hourly O₃ concentrations from the observational and model-derived data were compared to evaluate the performance of the model. The observed data were retrieved from the Atmospheric Environmental Regional Observation System (AEROS) of the Ministry of Environment of Japan. The modelled data were extracted from a grid cell containing the selected location. The hourly values for the entire period and hourly averages were compared. Pearson correlations and the root mean square error (RMSE) were calculated to evaluate the agreement between the observations and modeled data. Seven monitoring stations dispersed in the Kanto region were chosen for data comparison. The selected stations were the capital cities of Yokohama, Chiba, Shinjuku, Saitama, Mito, Maebashi, and Utsunomiya. These seven points are distributed in the urban and suburban areas of the Kanto Plain.

2.8. Ozone Responses to the Reduction Scenarios

Scientific evidence shows that the adverse respiratory effects of O₃ result from 6 to 8-h exposures [37]. Therefore, most of the epidemiological studies that evaluate the risk of exposure use the 8-h average for assessment [38]. Based on these data, this study calculated the O₃ distribution map and the O₃ differences between the base year and reduced scenarios for the 8 h average concentrations (10:00–18:00) during the O₃ season (April to September) 2016. In addition to the O₃ distributions and differences, the hourly averages for the O₃ season were compared in the seven capital cities of Kanto.

2.9. Benefit Evaluation for the Reduction Scenarios

The benefits of the simulated scenarios were evaluated in terms of the monetary impact that changes in O₃ concentrations have on human health and rice production. Benefits Evaluation System (BESystem) software was used for assessment. The BESystem software [39] was developed by the National Institute of Advanced Industrial Science and Technology of Japan (AIST) and considers changes in the number of premature deaths due to exposure to O₃ and estimates its monetary impact. To calculate the reductions in mortality rates, the BESystem uses the following equation:

$$\Delta Y= -Y_0\left(1-e^{-\beta \left(\Delta {\mathrm{O}}_3\right)}\right)$$

where Y₀ is the baseline incidence rate, β is the rate of increase in early deaths per 1 ppb O₃ increase in the 8 h average ozone, and ΔO₃ is the O₃ difference between the base year and the reduced scenario. The function was derived from the equation developed by the US Environmental Protection Agency in the Environmental Benefits Mapping and Analysis Program [40]. To calculate Y₀, BESystems uses information on the mortality rates (all causes, except accidental deaths) from the Ministry of Health, Labour and Welfare [41], classified by age and gender for all the municipalities in the simulated domain. The mortality rates are then interpolated with the population distribution in the municipalities using data from the Bureau of Statistics of Japan [42]. For β, the value established by Turner et al. was used [43]. The software considers the population distribution in every cell of the simulated domain, and to quantify the economic benefits of mortality changes, ΔY was multiplied by a valuation estimated by the value of statistical life (VSL). VSL estimates a person’s willingness to pay for a reduction in mortality risk. Table S2 lists the values of β and VSL in this study.

Because rice is an essential crop in Japan, the BESystem considers the adverse effects of O₃ on crop yields. Several studies have shown a relationship between crop yield and tropospheric O₃ concentrations [44,45]. The BESystem calculates the reduction in rice production due to O₃ with the following equation:

$$\Delta W= -W_0\times \gamma \left({\mathrm{O}}_3\right)\times \Delta {\mathrm{O}}_3$$

where W₀ is the baseline rice production, γ(O₃) is the rate of rice production reduction per 1 ppb increase in the 8 h averaged O₃ concentration, and ΔO₃ is the O₃ difference between the base year and the reduced scenario. The equation assumes a linear relationship between the ozone concentrations and rice yields based on the evidence from other studies [18]. For calculating W₀, crop yield data for the municipalities in the simulated domain were obtained from the Ministry of Agriculture, Forestry and Fisheries [46]. Thereafter, the rice crop yields are allocated in the domain by interpolation with land use data from the Bureau of Statistics.

To calculate the economic impacts of changes in rice production, ΔW was multiplied by the price of a kilogram of rice (V.W.). Table S2 lists the γ(O₃) and V.W. values used in this study.

3. Results and Discussion

3.1. Comparison of Observed and Modeled Data

Hourly O₃ values during the BASE scenario for the observed and simulated data are shown in Figure 2. The correlation factor (R) and root mean square error (RMSE) were calculated to evaluate the agreement between both datasets for each of the seven points. Yokohama and Maebashi showed the lowest R and highest RMSE, likely because the model underestimated the O₃ concentrations during the first days of the simulation at these two points. The RMSE indicated good agreement between the simulated and observed data for the other five points. The model correctly reproduced the patterns in the observed data, such as diurnal variations and high levels of pollution episodes that occurred on different days. However, the correlation factors showed better data agreement in the most urbanized areas of Shinjuku and Saitama and a tendency to overestimate the O₃ levels in the suburban areas.

The agreement in the dataset improved when comparing the hourly average for the O₃ season, as shown in Figure 3. The correlation factors ranked from 0.8 to 0.96, and the RMSE was also reduced. Similar to the hourly data, the concentrations in Mito, Maebashi, and Utsunomiya (suburban areas) were overestimated by the model, and the more urbanized areas of Shinjuku and Saitama displayed lower RMSE, indicating better data agreement. Nevertheless, the data comparison showed that the model effectively reproduced the O₃ concentrations at the compared points.

3.2. Ozone Distribution and Ozone Differences in the Kanto Region

Figure 4 shows the 8 h (10:00 to 18:00) average O₃ concentrations in Kanto for the BASE scenario during the O₃ season. Higher O₃ levels were observed in the western and northwestern Kanto region than those in the central region. Most of the areas with higher O₃ levels were suburban sites (Maebashi, Utsunomiya, and Mito) with lower precursor emissions (Figures S2 and S3). In contrast, areas with higher precursor emissions (i.e., Tokyo and Yokohama) displayed lower O₃ levels.

Although precursor emissions are a determining factor for O₃ concentrations in a specific area, the contribution of meteorology is equally important. Khiem et al. [18] showed that O₃ variations in Kanto are highly correlated with temperature and wind speed changes during the summer. Additionally, Kiriyama et al. [19] proved that pollutants emitted along Tokyo Bay are transported by the sea breeze to inland areas in the afternoon, leading to higher O₃ concentrations than those in the urban core. Thus, meteorological patterns contributed to the O₃ concentration distributions observed in Figure 4.

The O₃ differences between the reduction scenarios and the base year are shown in Figure 5. The maps were arranged as a function of the percentages of NO_x and VOC. There is marked heterogeneity in O₃ responses between the suburban and urban areas of Kanto. The suburban regions showed a decreasing tendency in most of the scenarios, and major decreases were found when NO_x emissions were reduced by >50%. Areas with decreased O₃ levels expanded as NO_x emissions reductions increased. Conversely, the urban regions showed an increasing pattern when NO_x emissions were reduced by >25%, and the increasing trend peaked when high NO_x reductions were mixed with no VOC level changes. For example, when NO_x emissions were reduced by 50% with no reduction in the VOC emissions, an increase of >9.5 ppb was observed in the urban core of Kanto and an increase between 6.5 and 9.5 ppb was observed in the rest of the Greater Tokyo Area.

However, according to Figure 5, the most effective way to achieve a generalized reduction in O₃ concentrations in the Kanto Plain is to significantly reduce the NO_x emissions in the entire region. As observed, when NO_x emissions were reduced by 75%, reductions up to 9.5 ppb were achieved in large areas of Kanto. Furthermore, the areas with decreased O₃ reached the urban core of Kanto as the NO_x emissions were decreased by 90%, and maximum O₃ reductions were observed when NO_x was reduced by 100%.

The spatial differences in O₃ changes conform to the O₃ sensitivity. Other studies have determined that O₃ formation in Kanto is unevenly distributed between the NO_x-limited and VOC-limited areas. Inoue et al. [25] proved that there is heterogeneity in O₃ formation sensitivity in the main urban regions of Japan, including the Kanto region. This suggests that the approach for effectively reducing tropospheric O₃ should be specific to every region. Under the current situation, O₃ concentrations in central Kanto must be approached from a VOC-limited perspective, and suburban areas must be approached from a NO_x-limited perspective. Therefore, significant reductions in VOC emissions, with few or no reductions in NO_x, would be an effective strategy to reduce O₃ levels in the short-term (Figures S4 and S5). Other studies have shown that combining significant reductions in VOC emissions with fewer changes in NO_x emissions is an effective strategy for reducing O₃ levels in VOC-limited areas [47,48].

Nonetheless, future policies in Japan aim to completely reduce NO_x emissions from anthropogenic sources by including new technologies, such as zero-emission vehicles, or more complex approaches, such as nitrogen cycling technology. Technologies for reducing NO_x emissions currently focus on selective and non-selective catalytic reduction. More efficient methods are being developed to help achieve extraordinary reductions in the NO_x levels. Substituting the current car fleet in the Kanto area with hybrid and zero-emission vehicles can reduce NO_x emissions by 20% and impact ozone concentrations [23]. Another promising technology considers reducing NO into ammonia (NH₃), a compound widely used as a fertilizer and potential source of energy. Although this technology is still under development and a detailed plan for its implementation is still under consideration, reduction efficiencies of 50% have been reported [49,50].

Therefore, in the coming years, the transition from a VOC-limited to a NO_x-limited regime is most likely to occur in Kanto owing to significant reductions in NO_x concentrations, which in turn influence the O₃ formation sensitivity. The evaluated scenarios focused on reducing anthropogenic VOC emissions. However, the model considers the biogenic VOC (BVOC) emissions. The reactivity and seasonal variability of terpenoids (important species in the BVOC emissions) might be more significant in the coming years due to rising temperatures [51]. Higher VOC emissions will shift ozone sensitivity to stronger NO_x-limited conditions. Still, under stronger NO_x-limited conditions, the reductions in the NO_x emissions proposed in this study will effectively reduce tropospheric ozone levels.

Although the N010 and N000 scenarios might be the most challenging to achieve, they exemplify the circumstances of future years if the net-zero emissions goal is to be reached by 2050. By significantly reducing the emissions of precursors, O₃ concentrations would also have a significant impact on minimizing the undesired effects of high O₃ concentrations and lowering the concentrations of greenhouse gases in the atmosphere. This approach is necessary to reduce ground O₃ levels in other urban areas as well. To effectively reduce O₃ levels in the Pearl River Data in China, significant reductions in NO_x emissions followed by a transition to the NOx-limited regime are needed to attain the AQS for ozone in China [11].

Background concentrations of pollutants are an important source of uncertainty in air quality simulations. The physical and chemical phenomena influencing background concentrations occur all the time and are not limited to a particular region. The performance of several models is influenced by the background ozone levels [52], and the bias might be greater when simulating future scenarios with significant changes in the precursor emissions, as in this study. This bias should be addressed for future studies to improve the model performance on ozone concentrations.

3.3. Concentration Differences in Seven Measuring Points in Kanto

Seven measuring points in each prefecture’s capital city were compared to evaluate O₃ changes at specific points in Kanto.

Table 2. Mean, maximum, and % difference in the ozone concentrations for the 21 simulated scenarios in seven points in Kanto (ppb). Values in red indicate increased ozone concentrations.

	Base	N100V050	N100V000	N075V100	N075V050	N075V000	N050V100	N050V050
Yokohama
Mean	21.61	21.03	20.46	26.51	25.86	25.21	31.69	30.99
Max	100.42	95.94	90.96	97.17	94.36	90.79	85.39	84.26
% diff mean		−2.69	−5.32	22.69	19.68	16.68	46.63	43.41
% diff max		−4.46	−9.42	−3.24	−6.03	−9.59	−14.97	−16.09
Chiba
Mean	33.25	32.86	32.45	34.94	34.57	34.17	36.01	35.73
Max	68.33	63.22	60.80	69.75	64.86	59.27	66.30	62.73
% diff mean		−1.18	−2.41	5.07	3.98	2.76	8.30	7.45
% diff max		−7.47	−11.02	2.09	−5.07	−13.26	−2.96	−8.19
Shinjuku
Mean	22.14	21.36	20.62	27.56	26.62	25.71	33.10	32.07
Max	84.35	81.56	78.27	78.69	76.93	74.52	85.88	72.67
% diff mean		−3.52	−6.85	24.48	20.23	16.14	49.53	44.88
% diff max		−3.31	−7.20	−6.71	−8.80	−11.66	1.81	−13.85
Saitama
Mean	29.10	27.78	26.59	33.79	32.16	30.65	38.10	36.59
Max	81.63	80.56	79.07	101.73	77.05	71.26	117.67	100.15
% diff mean		−4.54	−8.63	16.12	10.54	5.34	30.94	25.76
% diff max		−1.31	−3.14	24.62	−5.62	−12.71	44.15	22.69
Mito
Mean	39.35	38.96	38.51	38.04	37.92	39.11	38.84	38.51
Max	86.42	83.69	79.85	66.10	65.74	78.64	77.17	75.20
% diff mean		−1.00	−2.13	−3.32	−3.62	−0.59	−1.28	−2.12
% diff max		−3.17	−7.61	−23.52	−23.94	−9.01	−10.70	−12.99
Maebashi
Mean	36.82	34.69	32.55	38.87	37.24	35.15	38.95	38.00
Max	129.00	106.50	85.96	129.59	118.75	100.52	110.99	107.62
% diff mean		−5.79	−11.60	5.55	1.12	−4.54	5.77	3.19
% diff max		−17.44	−33.36	0.45	−7.95	−22.08	−13.96	−16.58
Utsunomiya
Mean	39.60	38.51	37.30	40.37	39.59	38.61	39.69	39.27
Max	104.11	96.92	87.56	97.53	94.14	89.06	83.01	81.80
% diff mean		−2.74	−5.82	1.95	−0.04	−2.50	0.23	−0.83
% diff max		−6.91	−15.90	−6.32	−9.58	−14.46	−20.26	−21.43
	N050V000	N025V100	N025V050		N025V000	N000V100	N000V050	N000V000
Yokohama
Mean	30.27	35.69	35.15		34.46	32.13	32.15	32.16
Max	82.76	61.83	61.95		61.97	51.58	51.58	51.58
% diff mean	40.10	65.16	62.65		59.46	48.70	48.78	48.84
% diff max	−17.59	−38.43	−38.31		−38.30	−48.64	−48.63	−48.63
Chiba
Mean	35.39	35.66	35.57		35.42	31.92	31.93	31.94
Max	58.32	51.89	51.01		50.46	52.18	52.17	52.17
% diff mean	6.43	7.25	6.97		6.52	−4.00	−3.97	−3.95
% diff max	−14.65	−24.05	−25.34		−26.15	−23.63	−23.64	−23.65
Shinjuku
Mean	30.97	37.06	36.44		35.53	33.21	33.23	33.24
Max	65.92	77.71	73.81		65.10	53.40	53.40	53.41
% diff mean	39.89	67.40	64.58		60.50	50.01	50.08	50.15
% diff max	−21.85	−7.87	−12.50		−22.82	−36.70	−36.69	−36.68
Saitama
Mean	34.72	39.51	38.90		37.82	32.62	32.64	32.66
Max	71.05	90.49	88.63		80.66	53.32	53.33	53.33
% diff mean	19.34	35.81	33.69		29.98	12.09	12.17	12.23
% diff max	−12.96	10.85	8.57		−1.19	−34.68	−34.67	−34.67
Mito
Mean	37.76	35.65	35.67		35.68	31.57	31.57	31.58
Max	65.12	52.87	52.36		51.69	52.18	52.18	52.18
% diff mean	−4.04	−9.39	−9.33		−9.31	−19.77	−19.76	−19.75
% diff max	−24.65	−38.82	−39.41		−40.19	−39.63	−39.63	−39.63
Maebashi
Mean	36.61	35.84	35.60		35.10	25.28	25.31	25.35
Max	100.97	78.97	78.81		77.64	50.08	50.09	50.11
% diff mean	−0.58	−2.67	−3.31		−4.67	−31.36	−31.26	−31.17
% diff max	−21.73	−38.78	−38.91		−39.81	−61.18	−61.17	−61.16
Utsunomiya
Mean	38.68	29.59	29.61		29.62	36.63	36.59	36.47
Max	79.77	48.88	48.89		48.88	59.58	59.81	59.74
% diff mean	−2.33	−25.28	−25.24		−25.20	−7.50	−7.60	−7.91
% diff max	−23.38	−53.05	−53.04		−53.05	−42.77	−42.55	−42.62

presents the parametric statistics for the seven analyzed points in the 21 simulated scenarios. The mean, maximum, and % differences between them between the base year and every reduced scenario are shown.

The points in the urban core (Yokohama, Shinjuku, and Saitama) only showed reductions when VOC emissions were reduced without modifying NO_x emissions (N100V050, N100V000). Reductions between 1% and 9% of the mean values were observed. However, when NO_x emissions were reduced by >10%, the mean O₃ levels increased. In VOC-limited areas, O₃ production efficiency increased as NO_x concentrations decreased. In addition, O₃ titration by NO (NO + O₃ → NO₂ +O₂) is an important process that influences the daily patterns of O₃ concentration in VOC-limited areas. Therefore, O₃ depletion through titration decreases when NOx emissions are reduced, causing an increase in ground O₃ levels.

In the suburban areas of Mito, Maebashi, and Utsunomiya, the mean values in the BASE scenario were markedly higher than those of the urban sites (Table 2). The urban cores of cities frequently register higher O₃ concentrations than suburban areas within the same city [53,54]. In suburban areas, the mix of higher biogenic VOC emissions, lower O₃ titration by NO, and the transport of O₃ precursors and O₃ from urban areas leads to higher tropospheric O₃ concentrations than in urban cores. Notably, the suburban points achieved reductions in the O₃ levels in most of the evaluated scenarios. However, the decrease is more significant when NO_x reductions are at their maximum (N010, N000). For Mito, Maebashi, and Utsunomiya, the reductions achieved when reducing the NO_x emissions to zero were significantly different from the base year, ranging between 20 to 31%. In suburban areas, when reducing the NO_x emissions to zero, O₃ concentrations decrease to background levels of approximately 30 ppb [55], which is considerably lower than the base year values, where concentrations >60 ppb are observed.

To analyze the hourly patterns, the average O₃ concentrations for scenarios with the same NO_x reduction and different VOC changes were calculated for the O₃ season (i.e., hourly averages of N090V100, N090V050, and N090V000). The results are shown in Figure 6. These trends agree with the descriptive statistics presented in Table 2. The patterns can be divided into urban (Yokohama, Chiba, Shinjuku, and Saitama) and suburban areas (Mito, Maebashi, and Utsunomiya). In the urban areas, O₃ concentrations decreased in the scenarios where NO_x emissions were not reduced (N100) or were reduced by 100% (N000). The N090 scenario also showed a slight O₃ reduction, which was more significant in Saitama than in Shinjuku and Yokohama. These results indicate that VOC emissions at these sites limit O₃ formation.

In contrast, in suburban regions, O₃ concentrations were reduced in all evaluated scenarios. However, major reductions were observed when the NO_x emissions were significantly reduced (N000, N010, and N025). Therefore, the patterns observed in Figure 6 confirm that a generalized decrease in O₃ concentrations in Kanto could be achieved if NO_x emissions were significantly reduced.

3.4. Benefit Evaluation for the Reduction Scenarios

The benefit calculation results measured as the economic changes between the base year and reduction scenarios are shown in

Table 3. Benefit analysis of the simulated scenarios. The benefit is measured as the monetary change in millions of yen between the base year and the reduction scenario.

Benefits	Scenarios
	N100V050	N100V000	N090V100	N090V050	N090V000	N075V100	N075V050
Health	181,148	356,842	−185,311	462	183,301	−669,258	−447,229
Crops	9456	1904	−23	903	1864	1989	1050
Total	182,094	358,747	−185,334	1365	185,165	−649,361	−447,018
	N075V000	N050V100	N050V050	N050V000	N025V100	N025V050	N025V000
Health	−259,722	−757,873	−603,140	−411,804	−662,268	−594,634	−483,012
Crops	211	1771	2266	2991	5691	5777	6001
Total	−258,672	−756,103	−600,874	−408,813	−656,577	−588,857	−477,011
	N010V100	N010V050	N010V000	N000V100	N000V050	N000V000
Health	−122,021	−119,300	−105,458	648,217	644,702	641,494
Crops	10,064	9980	9890	14,089	14,071	14,054
Total	−111,957	−109,320	−95,567	662,307	658,774	655,548

. Positive values indicate an increase in monetary value, whereas negative values (shown in red) indicate a decrease. The results were divided into health, crop, and total benefits. Seven of the 20 scenarios showed total positive benefits. The remaining 13 had a negative impact. The results in Table 3 agree with the O₃ differences observed in Figure 5; the scenarios achieving a positive economic change were those that achieved a generalized decrease in O₃ concentrations in the Kanto region.

Figure 7 and Figure 8 show the economic change divided into health and crop benefits for all ranges of NO_x reduction in two of the evaluated VOC reductions (V100 and V000). When mixing small NO_x reductions with significant VOC reductions (V100), the health benefits were positive. Subsequently, a transition to negative impacts was observed when NO_x reductions were >25% and was sustained until reductions were at 75%. Subsequently, when NO_x emissions were reduced by >90%, the economic change was positive again. The benefits achieved when NO_x was reduced by >90% were independent of the VOC reductions; this implies that under significant reductions in the NO_x emissions, changes in the VOC are not crucial for achieving a positive impact on the health benefits. The achieved economic change when NO_x was reduced by >90% is more significant than when NO_x was not reduced. However, studies evaluating the mortality risk associated with long-term exposure to O₃ [56] report significant uncertainties that should be considered when selecting a good strategy based on the health impacts of O₃ reduction.

Moreover, the benefits related to rice production showed a direct relationship with the NO_x cuts, increasing as the NO_x reductions became more significant. The benefits were the highest when NO_x reductions were at maximum (>75%); from this point, VOC emissions reductions did not influence the economic change. When reductions in NO_x emissions were <25%, the economic impacts remained unchanged in each VOC scenario (V000 and V100). Nonetheless, entirely reducing VOC emissions (V000) mixed with NO_x emissions <25% has greater economic benefit than that when VOC are not reduced (V100).

These observed patterns in the evaluation of the benefits were because the suburban areas achieved reductions in O₃ concentrations in most of the simulated scenarios, and the rice fields in Kanto are located in suburban regions. In contrast, the population is concentrated in the central Kanto region (Figure S6). This area experienced major increases in O₃ levels in several scenarios, causing a negative economic impact due to the increased risk of exposure to higher O₃ levels.

Therefore, 90% and 100% reductions in NO_x emissions, independent of the VOC reductions applied, had the most positive economic impact on health and rice production. Future implementation of these scenarios would improve economic benefits, reduce ground-level O₃, and achieve net zero greenhouse gas emissions in the Kanto region.

4. Conclusions

The heterogeneity in the O₃ responses to changes in the emission rates of precursors showed that specific O₃ countermeasures for each region must be applied. Some of the evaluated scenarios reduced the O₃ levels in the suburban areas of Kanto but increased them in the urban core. Consequently, there are two approaches for effectively reducing tropospheric O₃ concentrations: significant reductions in VOC emissions without changes in NO_x concentrations or significantly decreased NO_x levels. The latter strategy was the most effective for a generalized reduction in ground O₃ levels, positively impacting the health benefits and rice production. Therefore, this study showed that future Japanese technologies to significantly reduce NO_x emissions are crucial for achieving reductions in tropospheric O₃ levels in the region.